摘要
设计了一套基于机器视觉的大米外观品质参数检测装置,实现了对垩白度、垩白粒率、黄粒米和粒型参数的检测。该系统基于嵌入式计算机系统,应用改进的流域分割算法实现了粘连籽粒图像的分割,应用BP神经网络实现了垩白米的检测,应用色度实现了黄粒米的检测,应用极坐标下的长短轴快速检测算法实现了粒型的检测。试验结果表明,该装置对垩白粒率的检测精度为±2%,垩白度的检测精度为±1%;对黄粒米的检测精度为±5%;粒型的检测精度为±4%。
An instrument, based on embedded computer system, is presented in this paper, which was designed for rice appearance-quality detection. A new watershed algorithm improved from priori knowledge was adopted to kernel image segmentation. BP neural network was selected to detect rice kenrnel chalkiness, the chroma to detect yellow-coloured rice, and a polar-coordinates-based algorithm to detect the kernel shape by measuring the long and short axis of each region. The experimental results show that the detection accuracy of chalky rice numbers, chalk degree, yellow-coloured rice, and rice kernel shape are ±2%, ±1%, ±5%, and ±4% respectively.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2005年第9期89-92,共4页
Transactions of the Chinese Society for Agricultural Machinery
基金
"九五"国家科技攻关计划资助项目(项目编号:990100112)
关键词
大米
外观品质
机器视觉
图像处理
Rice, Appearance quality, Machine vision, Image processing
